Hydrogen-bonding parameter and its significance in quantitative

mydriasis test for anticholinergic activity, and the procedure for assessment of H2-receptor inhi- bitory activity in vitro. The effect of 5b on the r...
0 downloads 0 Views 1MB Size
Hydrogen-Bonding Parameter

mp 142-143.5 “C; UV max (95% EtOH) 243.5 nm (t 13820);IR (KBr) 1672 cm-’; NMR (CDC13)6 1.30 (3 H, s), 1.37 (3 H, s), 1.57 (3 H, s), 1.97 (3 H, s), 2.00 (3 H, s), 2.23 (1H, d, J = 13 Hz), 2.73 (1 H, d, J = 13 Hz), 6.0-6.6 (2 H, br), 7.0-7.8 (5 H, m). Biological Methods. The following test procedures have been described previously:2 the 5-h pyloric-ligated rat test, the method for measuring inhibition of histamine-induced gastric acid secretion in the dog, the mouse mydriasis test for anticholinergic activity, and the procedure for assessment of H2-receptorinhibitory activity in vitro. The effect of 5b on the response of guinea pig auricles to carbamylcholine was measured in vitro in a Krebs-Henseleit solution at 35 “C. The compound was dissolved in Me2SO-H20 and then added to the bath solution. The tissue was exposed to the drug for 5 min. A t concentrationsof lo4 and M the compound did not affect the dose-related carbamylcholine-induced decrease in spontaneous heart rate. At higher concentrations the drug precipitated from the bath solution. In a similar manner the effect of 5b on the response of guinea pig trachea to carbamylcholinewas measured. At a concentration of M the drug did not affect the dose-related carbamyl-

Journal of Medicinal Chemistry, 1977, Vol. 20, No. 8

1071

choline-induced constriction of the trachea.

References and Notes (1) Presented at the Seventh Northeast Regional Meeting of the American Chemical Society, Albany, N.Y., 1976. (2) M. R. Bell, A. W. Zalay, R. Oesterlin, S. D. Clemans, D. J. Dumas, J. C. Bradford, and J. Rozitis, Jr., J . Med. Chem., 20, 537 (1977). (3) P. Bass, Adv. Drug Res., 8, 209 (1974). (4) B. R. Brodie, J. R. Gilette, and B. N. LaDu, Annu. Rev. Biochem., 27, 427 (1958). (5) A. Treibs, K. Jacob, F.-H.Kreuzer, and R. Tribollet, Justus Liebigs Ann. Chem., 742, 107 (1970). (6) R. W. Brimblecombe, W. A. M. Duncan, G. J. Durant, J. C. Emmett, C. R. Ganellin, and M. E. Parsons, J . Znt. Med. Res., 3, 86 (1975). (7) We are indebted to Dr. Bertram A. Spilker and his associates for these results. (8) J. W. Black, W. A. M. Duncan, J. C. Emmett, T. Hesselbo, M. E. Parsons, and J. H. Wyllie, Agents Actions, 3, 133 (1973).

Hydrogen-Bonding Parameter and Its Significance in Quantitative Struct ure-Activity Studies Toshio Fujita,’ Takaaki Nishioka, and Minoru Nakajima Department of Agricultural Chemistry, Kyoto University, Kyoto, Japan 606. Received September 27, 1976

When the relative hydrogen-bondingeffect of drugs on phases involved in the binding at the site of biological action differs from that in the l-octanol-H20 partitioning phases used as the reference to estimate the hydrophobicity, a parameter (or parameters) which represents the “extra” hydrogen-bondingeffect on the biological activity is required in the Hansch-type correlations. As a first approximation, the effect is analyzed in terms of the ratio of hydrogen-bondingassociation constants and the ratio of molarities of hydrogen-bondingspecies constituting the biological and organic phases. Sometimes, the association constants in both phases are so similar that they are not important in determining the extra hydrogen-bondingeffect. The net result is that the effect is expressible by an indicator variable term the slope of which corresponds to the molarity ratio. The variable only applies to substituentshaving appreciable association capability in correlating a certain biological action exhibited by a series of congeners. Quantitative analysis of structureactivity relationships of the Hansch-type assumes that the variation in a certain biological activity expressed as A log BR of a series of congeners can be analyzed in terms of variations in their free-energy-related physicochemical parameters, such as ?r or log P for hydrophobic, u, u*, or A log KAfor electronic, and E,, E:, or MV for steric and others.’ Although hydrogen bonding has long been recognized as being very important in biological reactions including drug actions, examples where the particular use of hydrogen-bonding parameters as an independent variable is necessary have scarcely been found so far. One of the reasons for such scarcity may stand on the fact that the hydrogen-bonding effect at a certain functional group in congeners is correlated with the electronic effect of substituents. Examples are found in analyses for the Hill reaction inhibition of herbicidal N-substituted phenylamides’ and the bacteriostatic activity of kojic acid analogues3 where hydrogen-bonding effects of such functional groupings as -NHCO- and -COC(OH)= on cellular components are anticipated. In fact, the freeenergy-related hydrogen bond donating parameter, hD, defined by Higuchi and co-workers4 for a series of substituted phenols can be expressed by eq 1. A similar observation was made by Clotman and co-workers for the association equilibrium of substituted phenols with trieth~lamine.~ Likewise, the hydrogen-accepting parameter, ~KHB of, substituted benzaldehydes has been expressed

as eq 2 by Taft and associates.6 uc correlates better than u since the site of association is partially electron deficient. h D = 1.562 (k0.169) u + 0.096 (k0.064) (1) n = 22; r = 0.974; s = 0.115

PK- = -0.456 (kO.101) d + 0.782 (k0.092) (2) n = 5; r = 0.993; s = 0.061 Another line of reasoning may be that the parameter used for the hydrophobicity, log P or K, mostly estimated from the l-octanol-H20 partition coefficient, inherently includes the effect of hydrogen bonding. According to Hansch and co-workers, the nonspecific type of protein binding of miscellaneous molecules such as aromatic hydrocarbons, phenols, anilines, and aliphatic alcohols with bovine serum albumin (BSA)7 and bovine hemoglobin (BHG)’ is related only to their hydrophobicity defined by the octanol-HzO partition coefficient. The binding equilibrium constant, log K (K is the reciprocal of the concentration of compound producing a 1:l complex with protein), is linearly related to log P (octanol-H20) without any additional parameter, regardless of the number of hydrogen bonding sites of varying degrees of strength in the molecule as shown in eq 3 and 4. The effect of the relative hydrogen bonding on the odanol-HzO partitioning is closely similar to that involved in the protein binding. Thus, the net difference in the hydrogen-bonding effect between the partitioning process and drug actions, whose

1072 Journal of Medicinal Chemistry, 1977, Vol. 20, No. 8

Fujita et al.

critical step occurs rather nonspecifidy in the proteinous milieu, does not necessarily appear as an explanatory independent parameter in the Hansch-type correlations.

For BSA binding’ log K = 0.751 (k0.07) log P + 2.301 (20.15) n = 42; r = 0.960; s = 0.159

(3)

For BHG binding’

logK = 0.713 ( i O . 1 1 ) l o g P + 1.512 (k0.27) n = 17; r = 0.950; s = 0.160 log l / p = 1.17 (t0.25) lOgP (Oct-HZO) + 1.88 (k0.33) I - 2.11 (i0.39) n = 30; r = 0.947; s = 0.438 log l/Md= 1.55 (t0.26) 7~ - 1.93 (k0.52) uo + 1.53 (k0.36) HB + 2.46 (t0.28) n = 21; r = 0.967; s = 0.247

(4)

(5)

(6)

However, examples such as eq 5 and 6 where hydrogen-bonding effects appear to be expressible by an indicator variable have been found quite recently. In eq 5 derived by Hansch and co-workers: p is the effective anesthetic pressure of gaseous anesthetics and I is an indicator variable which takes a value of 1 when aliphatic anesthetics have an activated hydrogen atom attached directly to a carbon atom holding electronegative substituentts), but otherwise 0. Equation 6 is derived from our acetylcholinesterase (AcChE) inhibition data of meta-substituted phenyl N-methylcarbamates.” Kd is the dissociation constant of the enzyme-inhibitor reversible complex and HB, an indicator variable, takes 1 only for hydrogen-acceptor substituents such as NOz, CN, CHO, Ac, COEt, and N(Me),. uo which is almost identical with G for meta substituents except for alkoxy groups is used in conformity with correlations for ortho and para isomers where uo works better than a-.lo It may be difficult to accept that the hydrogen-bonding capability of activated hydrogens in eq 5 and hydrogen accepting groups in eq 6 having “various” degrees of acidity and basicity can be represented by a discrete-type parameter. We attempted to clarify whether physicochemical significance can be rendered to this “hydrogen bonding” parameter. Although log P or ?r values from the octanol-H20 partitioning system may take care of the nonspecific type of binding process with cellular components including both the hydrophobic and hydrogen-bonding interactions, the relative hydrogen-bonding capability of drugs between octanol and water phases may differ from those between phases involved in such biological actions as analyzed by eq 5 and 6. We examined relationships between log P values determined with various solvent systems differing in the relative hydrogen-bonding capability for series of congeners having substituents of varying degrees of hydrogen-bonding properties. This paper makes the point that the difference in hydrogenbonding effect of certain classes of substituents between certain different partitioning systems is, in effect, expressible by the indicator variable as a first approximation. It also discusses the scope and limitation for the use of the indicator variable as well as implications in cases where such variables appear in structure-activity studies.

1

2

3

Log POctOH

Figure 1. Plot of log P (CHC13-H20) vs. log P (octanol-H20) of monosubstituted benzenes. mg of each sample was shaken with 2-40 mL of CHC13and 2C-40 mL of H 2 0 a t 25 i 1 “C for 1 h. After separating two phases by centrifuging the whole for 20 min, the concentration in the water phase WEB measured spectrophotometrically as the corresponding phenols after alkaline hydrolysis according to the procedure described previously.” The concentration in the CHC13 phase was obtained by the difference. For monosubstituted benzenes, 50-400 mg of each sample was partitioned between 5 mL of CHC13 and 70 mL of H 2 0 by shaking for 1 h a t 25 i 1 “C. The concentrations in both phases were measured spectrophotometrically in this case. Determinations were performed a t least a t three different volume ratios or concentrations and the reproducibility was examined. Other partition coefficients used in this work were selected from literature. The relationships between log P values from various solvent systems were examined by means of regression analysis. The calculations were performed with a FACOM 230/75 computer a t the Data Processing Center of this University.

Experimental Section

Results Figure 1shows the relationship between log P values of monosubstituted benzenes from CHC13-H20 and octanol-H20 partitioning systems. The values for nonhydrogen bonders and hydrogen acceptors are related to each other by two parallel lines. Using the data in Table I, the relationship is formulated by eq 7 with HB as the hydrogen-bonding indicator variable. Hydrogen-acceptor substituents, such as alkoxy, NOz, CN, Ac, CON(Me)2, sulfonyl, and phosphoryl, seem to have almost equivalent effects to raise the log P (CHClS-HzO) value relative to that predicted from those of nonhydrogen bonders regardless of differences in the basicity. On the contrary, the effect of acidic (amphiprotic) substituents is to cause negative deviations from the regression line for nonhydrogen bonders to varying extents. log P (CHC13-HZO) = 0.801 (k0.087) lOgP (Oct-HZO) + 0.367 (k0.142) HB + 1.157 (t0.244) (7) rz = 14; r = 0.991; s = 0.081

P a r t i t i o n Coefficients. Partition coefficients with the CHC13-H20 system were determined for monosubstituted benzenes and phenyl N-methylcarbamates having neutral and hydrogen-accepting groups. For substituted carbamates, 20-60

log P (hex-H,O) = 0.804 (i0.338) log P (oct-H,O) + 0.718 (k0.881) (8) n = 5; r = 0.975; s = 0.079

Hydrogen-Bonding Parameter Table I. Log P (CHC1,-H,O) and Log P (Octanol-H,O) of Monosubstituted Benzenes Log P (CHC1,-H20) Substituents

Obsd'

H Me Et F Br

Nonhydrogen Bonders 2.80 2.86 0.06 2.13 3.41 3.31 0.10 2.69 3.6ae 3.68 0.00 3.15' 2.85 2.97 0.12 2.27 3.46e 3.43 0.03 2.84 3.61 3.55 0.06 2.99

OMe OEt Ac NO 2 CN CON( Me) , OPO( OMe), S0,Me

Hydrogen 3.12 3.21 3.62 3.53 2.79 2.79 2.93 3.01 2.71 2.77 2.01 2.02 2.56 2.50 2.00 1.90

c1

"2

CH2NH, NHAc NHCSNH, CONH, OH CH,COOH COOH

Calcdb

" ;:: H,O)' HB

IAI

Acceptors 0.09 2.11 0.09 2.51' 0.00 1.58 0.08 1.85 0.06 1.56 0.01 0.62' 0.06 1.22m 0.10 0.47n

Amphiprotic Substituentsf 1.428 1.88 0.46 0.90 l . l a h 2.03 0.85 1.09' 0.88g 2.09 1.21 1.16 0.54' 1.74 1.20 0.73' O.lli 1.67 1.56 0.64 0.36g 2.33 1.97 1.46 0.45' 2.29 1.84 1.41 0.468 2.64 2.18 1.81

0 0 0 0 0 0

3pKmd -0.36O -0.29O -0.34O -1.12'

-0.8oP -0.84'

1 1

1 1 1

1 1

1

0.02 0.22q 1.13 0.73 0.79 2.22 2.21r 1.05'

0" 0 : Nan-hydrogen

J

0:HydrOgen

Lmders

accei iors

00

0 0 0 0 0 0

oo

CHO

0 0

NMeAc

-1 COOH

Unless noted, data from this laboratory; the standard error (SE) is less than f0.02. By eq 7. ' Unless noted, from T. Fujita, J. Iwasa, and C. Hansch, J. A m . Chem. SOC.,86, 5175 (1964). Unless noted, from ref 6. e SE < *0.04. f Not included for t h e correlation. g C. Hogben, D. Tocco, and B. Brodie, J. Pharmacol. E x p . Ther., 125, 275 (1959). H. Smith, J. Phys. Chem., 25, 204 (1921). ' K. B. Sandell, Monatsh. Chem., 8 9 , 3 6 (1958). 1 H. Smith and T. White, J. Phys. Chem., 33, 1953 (1929). J. Iwasa, T. Fujita, and C. Hansch, J. Med. Chem., 8, 150 (1965). Unpublished data by Hansch and co-workers, from the log P data file of t h e Pomona College Medicinal Chemistry Project. K. C. Hansch Kamoshita and T. Fujita, unpublished data. and S. Anderson, J. Org. Chem., 32, 2583 (1967). O Estimated from the log K fvalue with PhOH: Z. Yoshida and E. Osawa, J. Am. Chem. Soc., 88, 4019 (1966). p Estimated from the log Kfvalue with MeOH: P. J. Krueger and H. D. Mettee, Can. J. Chem., 42, 288 (1964). Estimated from the log K fvalue with PhOH: T. Gramstad, Spectrochim. Acta, 1 9 , 4 9 7 (1963). Estimated from ~ K H ,values of (PhO),PO and (MeO),PO listed in ref 6 by interpolation. ' Estimated from the log K f value with PhOH: P. Biscarini, G. Galloni, and S. Gersetti, Spectrochim. Acta, 20, 267 (1964).

'

2-

'OctOH

Figure 2. Plot of log P (hexane-H20) vs. log P (octanol-H20) of monosubstituted benzenes. 0 : Non-hydrogen

0:

bonders

Hydrogen acceptors

'

The situation for log P (hexane-H20) values is depicted in Figure 2 using the data in Table 11. The regression line is drawn for nonhydrogen bonders for which eq 8 is obtained. As evident, the effect of hydrogen acceptor as well as amphiprotic substituents is not uniform lowering the log P (hexane-H20) value to various degrees. log P (CHCl,-H,O) = 0.949 (kO.111) log P (Oct-HZO) + 0.303 (k0.137 ) HB - 0.231 (k0.121) u

+ 0.731 (k0.239)

(9)

n = 24; r = 0.984; s = 0.096 For 24 ortho-, meta-, and para-substituted N-methylphenylcarbamates, eq 9 is derived from the data in Table 111. Equation 9 seems to separate the effect of variable substituents on the relative hydrogen-bonding solvation

I ik

31 -1

V

P

cn J

2t 11

2

1%1

1

1

Log POctOH

-

2 0.236

3

Figure 3. Plot of log P (CHC13-H20) vs. log P (octanol-H20) of substituted phenyl N-methylcarbamates.

at the side-chain carbamoyl groups, -OC(=O)NHCH3 (expressed by -0.23 u ) , and the hydrogen-accepting effect at the substituent site on the relative partitioning processes. For the electronic effect of ortho substituents, up was used. The effect of ortho substituents has recently been shown" to be expressible by a linear combination of up, the Swain-Lupton-Hansch F,13 and Taft-KutterThe fact that eq 9 does not require Hansch E, con~tanh.'~ F and E, terms suggests that the proximity electronic (or field inductive) and steric effects of ortho substituenh are not important in determining the relative solvation effect at the side chain. Figure 3 depicts the plot of log P (CHC13-H20)against the value of log P (octanol-H20) 0.23 Q. The 2-0-i-Pr derivative does not require the in-

1074 Journal of Medicinal Chemistry, 1977, Vol. 20, No. 8

Table 11. Log P (Hexane-H,O) and Log P (Octanol-H,O) of Monosubstituted Benzenes

Substituents

Log P (hexane-H,O) Obsda Calcdb IAI

H,O)C

pKmd -

H Me F

c1 Br

OMe N(Me), NO, CN CHO Ac CH,N(Me), NMeAc 2"

CH,NH, OH CH,OH COOH

Nonhydrogen Bonders 2.45 2.43 0.02 2.99 2.88 0.11 2.48 2.54 0.06 2.98 3.00 0.02 3.08 3.12 0.04 Hydrogen Acceptors 2.16e1f 2.41 0.25 2.38h 2.58 0.20 1.49 2.21 0.72 0.96e,f 1.97 1.01 l.llg 1 . 9 1 0.80 1.14e-g 1.99 0.85 1.44e,1 2.31 0.87 -0.40h 1.62 2.02 Amphiprotic Substituents -0.05 1.44 1.49 -0.15e,j 1.59 1.74 -0.89 1.89 2.78 -0.76h 1.60 2.36 -1.29 2.17 3.46

2.13 2.69 2.27 2.84 2.99 2.11 2.31 1.85 1.56 1.48f 1.58 1.98' 1.12m

-0.36' -0.29' -1.12p -0.8oP -0.84'

-

0.02 0.454 0.73 0.80 0.80 1.13 1.56 2.0'

0.90 1.09" 1.46 1.10 1.81

a Unless noted, from T. Sekine, Y. Suzuki, and N. Ihara, Bull. Chem. SOC.Jpn., 46, 995 (1973). For e 8, only By eq 8. the values of nonhydrogen bonders are used. Unless noted, from T. Fujita, J. Iwasa, and C. Hansch, J. A m . Chem. SOC., 86, 5175 (1964). Unless noted, from ref 6. e Estimated from the log P (heptane-H,O) value assuming that the P value in terms of mole fractions instead of molarities is equivalent to that from the hexaneH,O system. For relevancy of this assumption, see R. Aveyard and R. Mitchell, Trans. Faraday SOC.,65, 2645 (1969); 66, 37 (1970). Unpublished data by Hansch and co-workers, from the log P data file of the Pomona College Medicinal Chemistry Project. K . Nasim, M. C. Meyer, and J. Austin, J. Pharm. Sci., 61, 1775 (1972). H. L. Schmidt, M. Moller, and N. Weber, Biochem. Pharmacol., 22, 2989 (1973). A. Chao and G. Miwa, Drug Metab. Dispos., 2, 477 (1973). G. Williams and F. G. Soper, J. Chem. SOC.,2469 (1930). A. C. McCulloch and B. H. Stock, Aust. J. Pharm., 48, ,914 (1966). C. Takayama, T. Fujita, and M. Nakajima, unpublished data. W. J. Dunn, unpublished data. 3. Iwasa, T. Fujita, and C. Hansch, J . M e d . Chem., 8, 150 (1965). * See footnote o in Table I. P See footnote p in Table I. L. Joris, J. Mitsky, and R. W. Taft, J. A m . Chem. SOC.,94, 3438 (1972). Estimated from pK, values for N,N-disubstituted amides in ref 6.

Fujita et al. Table 111. Log P Values of Substituted Phenyl N-Meth ylcarbamates LogP (CHCL-H,O) (actSubstituents Obsda Calcdb H , O F 1

Table IV. Log P Values of Substituted Benzenesulfonamides Log P (CHCl,-H,O) Substituents H 3-Me 3-C1 3-N02 4-Me 4-OMe 4-C1 4-Br 4-AC 4-CN 4-NO: 4-NH2 4-"Me 3-CF ,-4-NO ~

dicator variable suggesting that the substituent is not basic enough, probably due to a steric congestion around the alkoxy lone pair electrons preventing the hydrogen hondine.

log P (CHC13-H20) = 0.890 (i0.202) lOgP (oct-H,O) - 1.015 (t0.428) u + 0.355 (k0.366) HB1 - 1.051 (k0.308) HB2 - 0.516 (k0.182) (10) n = 14;r = 0.988; s = 0.106 Equation 10 is derived from log P values in Table IV of meta- and para-substituted benzenesulfonamides determined by Kakeya and co-workers.'' In this equation, two indicator variables, HB1 for electron-accepting NO2, CN, and acetyl and HB2 for amphiprotic NH2 and NHMe groups, are introduced. Since collinearity between HB1 and (T in this set of congeners is somewhat high (the simple correlation coefficient = 0.786), and also since the data points contributing to the HB2term number only two, the

HB

o

.

H 1.78 1.83 1.16 0 0 2-1 2.55 2.55 0 1.96 0.18 2-OMe 1.81 1.86 1 0.81 -0.27 2-OEt 2.25 2.27 1.24 -0.24 1 2-0-i-Pr 2.23d 2.28 1.52 -0.45 0 2-CN 1 1.80 1.70 0.86 0.66 1 2-N02 1.86 1.82 1.02 0.78 3-Me 2.41 2.36 1.70 --0.07 0 3-i-Pr 3.37e 3.24 2.63 -0.07 0 3-C1 2.50 2.57 2.03 0.37 0 3-Br 2.81 2.78 2.25 0.39 0 3-OMe 2.33 2.24 1.30 0.12 1 3-NMe2 2.46d 2.44 1.43 --0.21 1 3-CHO 1.83 1 1.63 0.92 0.35 3-AC 1.93 1.80 0.90 0.38 1 3-CN 1.87 1.83 0.98 0.56 1 3-NO, 2.09 2.19 1.39 0.71 1 4-i-Pr 3.47 3.42 2.80 0.15 0 4-1 2.93 3.02 2.46 0.18 0 4-OEt 2.59 2.64 1.64 -0.24 1 4-CHO 1.88 1.89 0.99 0.42 1 4-AC 1.95 1.88 1.01 0.50 1 4-CN 1.88 1.78 0.95 0.66 1 4-N02 2.05 2.24 1.46 0.78 1 a The standard error (SE) is less than 50.02 unless noted. By eq 9. From ref 11. SE is i0.04. e SE is 20.08.

fty:

Obsd'

Calcdb H,O)a

u

-2.24 0.32 0.26 -0.36 0.33

0.31 -0.24 0.31 0.85 0.26 1.29 -0.39 0.55 0.39 0.82 0.47 0.18 0.00 0.84 0.46 1.36 -0.49 0.20 0.23 -0.63 0.64 -0.38 --1.64 -0.83 0.08 -0.64 0.15 1.73

0 -0.07 0.37 0.71

0.15

0.14 0.39 -0.36 -0.61 -0.60 -1.69 -0.59 0.19

HB, HB:

0 0 0 1 -0.17 0 --0.27 0 0.23 0 0.23 0 0.50 1 0.66 1 0.78 1 -0.66 0 -0.84 0 1.21 1

0 0 0 0 0 0 0 0 0 0 0 1 1 0

From re€ 15. From the original set, 3,4-CI, and 3-N02-4-C1derivatives are deleted. According to the private communication from the original authors, log P values of these compounds may not be very accurate due to a lack of enough solubility. By eq 10. a

data set might not really be adequate to what this type of analyses requires. However, the HB1 term is justified at the 94% level of significance and the slope of this term, 0.36, is quite close to those in eq 7 and 9. The difference in relative hydrogen-bonding effect between CHC13-H20 and octanol-H20 systems seems to be nearly constant for hydrogen-acceptor aromatic substituents, in three series of congeners being represented as 0.34 f 0.04 log unit. In eq 10, the OMe substituent is not classified as a hydrogen acceptor. A decrease in basicity of the oxygen lone pair electrons may be caused by an electron withdrawal from the strongly negative SO2"* group. The above results seem to indicate that the indicator variable applies in certain cases to the hydrogen-bonding effect of substituents on relative partitioning processes.

Journal of Medicinal Chemistry, 1977, Vol. 20, No. 8 1075

Hydrogen-Bonding Parameter

Discussion In the partitioning phases considered above, except for n-hexane, compounds having hydrogen-accepting substituents (X) are believed to exist as hydrogen-bonding equilibrium mixtures. For monosubstituted benzenes, the situation is shown as eq 11-13, where Kx is the association constant for substituent X and oct, CL, and W denote the solvents. Total concentration of the molecule in each + Kxmt[oct]),CocL(1+ KxcL[CL]), phase is given by C?mt(l and Cow(l + KxW[H2O])where Co is the concentration of the unassociated molecule and [solvent] is the molarity of the solvent itself in that solvent phase. The partition coefficients for hydrogen-acceptor molecules are then expressed as in eq 14 and 15. ArX

+

C,H,,OH

ArX

+

CHC1,

Kx°Ct

ArX, .HOC,H,,

(11)

cc, = COCL(1

+

+

KcoOCtK"OCt

log PCL = log

(20)

KcoCL[CL])

Coct = COOCt(l+ K,oO"t[oct]

+

K"OCt[OCt]

[octl')

Pact +

(21)

log (1 + K,OCL[CL])/

(1+ KcO°Ct[oct] + K"OCt[OCt]

+ ~ c o o C t ~ N H o C t [+~ constant ~t]2) log PCL

=

log

+

Pact +

log [CL]/[oct]

(23)

log KCOCL/KCOOCtK"OCt

+

log KXCL/I> ~ l ~1,] to eq 23.

1076 Journal of Medicinal Chemistry, 1977, Vol. 20, No. 8

log P,,

=

log Pact

+ log (1+ KyCL[CL])/

(1+ Ky°Ct[oct] f KHoct[oct]

+ KyOCtKHOCt[OCt]2) + constant (25) log Pc, log Pact + log K ~ ~ + log ~ l/KH°Ct / K + log [ CL] /[act]' + constant (26) 2

For log P (hexane-HzO) of monosubstituted benzenes having neutral, basic, and amphiprotic substituents, we can derive eq 27-29, respectively. log P,, = log Po,, + constant (27) log Phex= log Pact log I/(1 f Kx°Ct[ o c ~ ] ) i- constant log Poct+ log l/Kxoct

+ log l / [ o c t ] + constant

(28)

log Phex = log Pact log 1/( 1 + Ky°Ct[ O C ~ ] + KHoct[oct] K y ~ c t K H ~ ~ t [ o c t Iconstant 2)

+

+

+

log Poct log l/KyOCtKHOCt

+ log l/[oct]'

+ constant (29) In eq 26, 28, and 29, the hydrogen-bond formation constants only appearing as the denominator in respective terms are irreducible. Therefore, the effect of extra hydrogen bonding for these solvent-solute combinations is not represented by the indicator variable but by the negative deviations from eq 17 and 27 which vary according to the magnitude of KOct values. log P C , = log Pact + log KSOCL/K~HoCtKSOoCt + log [CL]/[oct]2 + log KyCL/KyOCt + log 1 / K p + log [CL]/[oct]2 + constant

(30) For benzenesulfonamides having amphiprotic substituents such as 4-NHz and 4-NHMe, eq 30 relates log P values from CHC13-H20 and octanol-HzO systems. The is log (KyCL/ difference from eq 23 (replaced by KSO'~'") KyOCt)+ log (l/KHOCt) log ([CL]/[OC~]~). Therefore, even though the term, log (KyCL/KyoCt), is reducible, the hydrogen-bonding effect should vary according to the magnitude of KHaCtfor a wide variety of amphiprotic substituents. The fact that the effect comes out only as the indicator variable term, 1.05HBz,in eq 10 means that the log KHOCtvalue does not differ so much between 4-"z and 4-"Me substituents. Since the value of log values are likely to be ([CL]/ [octI2)is -0.5, the log KHoCt in the order of 0.5. In fact, the hydrogen-bond formation constants of aniline as the hydrogen donor with oxy compounds are about 1.5-fold those of N-methylaniline in cyclohexane.lgb When this rather small difference (0.17 log unit) applies to the corresponding sulfonamide derivatives in the partitioning octanol solvent, it is not enough to discriminate between the 4-NHz and 4-"Me substituents for which the averaged total effect amounts to 1.05 log unit. According to Taft and his co-workers,6 the hydrogenbonding constant, Kf, of various bases with OH reference acids is expressed by eq 31 where PKHB is a constant representing the hydrogen-accepting ability of organic bases defined as the logarithm of the association constant with 4-fluorophenol in CC14 and m and c are constants characteristic of the reference acid and experimental conditions. We have estimated the PKHB value for some monosubstituted benzenes, whose Kf values with phenol or methanol are available in literature but not with 4-

+

Fujita et al.

fluorophenol, using eq 31 and m and c values for these reference acids.6 log Kf = mpK, + c (31) As shown in Tables I and 11, nonhydrogen bonders and ~ hydrogen ~ ~acceptors seem to be classified according to their PKHB value. For "nonhydrogen bonders", perhaps Kx[solv] > 1, the hydrogen-bonding effect on log PcLis almost uniform in spite of the wide variety in basicity, a range of 150-fold KHB units from anisole to Nfl-dimethylbenzamide. For log Phex,the effect of hydrogen bonding is such that the higher the pKHB value, the larger the downward deviation, A, from eq 8. The hydrogen-bond strength of hydrogen acceptors with CHC13 and octanol solvents was expected to be nearly equivalent to each other earlier in this paper. According to Taft and his associates,6 the association constants of various bases with methanol and ethanol are correlated by eq 31 where m = 0.5 and c = -0.4 to -0.5 at 25 "C. For l-octanol as the reference acid, a quite similar relationship is expected to hold under the same conditions. Gramstad and VikaneIgderived eq 32 from the association constants, Kf, of CHC13and phenol as hydrogen donors for a number ( n = 16) of hydrogen acceptors including P=O, C=O, and S=O compounds at 25 "C in CCl,. For phenol as the reference acid, m = 0.97 and c = -0.13 at 25 "C in eq 3L6 Therefore, the log Kf (phenol) values are almost equivalent to corresponding PKHB values. Thus, eq 32, where log Kf (phenol) was replaced by PKHB, is quite similar to correlations for alkanols as the reference acid. Under conditions of partitioning experiments where organic solvents as the hydrogen donor were water-saturated, the m and c values could be even closer between two systems. log ISf (CHC13)= 0.40 log Kf (phenol) - 0.85 ( 3 2 ) If we assume that eq 31 with m = 0.5 and c = -0.4 to 4 . 5 can be applied to the water-saturated octanol solvent as the reference acid, the log Kx°Ctvalue for anisole is around -0.5 from which the Kx°Ctvalue, ca. 0.3, is estimated. Since [CL] = 12.4 M and [oct] = 6.2 M, the conditions KXBo'V[s~l~] >> 1cannot apply to anisole when K OCt N KanisoleCL. Nevertheless, the value of log [(l+ KFPCL])/(l + KxWt[oct])]in eq 18 does not differ much from its reduced form, log ([CL]/[oct]). The values are 0.21 and 0.30, respectively, with the difference being 0.09 so that the hydrogen-bondingeffect of this compound can be approximately represented as the common indicator term with other hydrogen acceptors. Anisole seems to be the compound which is located around the borderline differentiating compounds whose HB = 0 and 1. The assignment, HB = 1,for the aromatic OMe substituent is no longer warrantable in disubstituted benzenes having strongly electron-withdrawing substituents besides OMe. The KOsfe50'V values may be even lower, Le., K O M ~ ~ ~ I " [ S O ~ V ] > 1, are roughly satisfied for water (55.5 M) and l-octanol (0.2-0.3) is about in the same order as that of alkanols (6.3 M). The extra term in eq 47 is reduced to log (KcHw/KcHoct) + log ( [ H ~ Q ] / [ o c ~$log ] ) (1 -t K c ~ ' ~ ' [ ~ i l ] ) (o_0.2),25the first term in the expression, log KXEnz/Kxmt log [Enz-H]/[oct], which is supposed to correspond to which becomes 1.40-1.50 when KCH = KcH"~,KCH'~'= the indicator variable, could be ignored. 0.5-0.8, and [oil] = 1.04 X 3 M counting three carbonyl Equation 52 correlates the log 1/Kd values for AcChE functions in the molecule. inhibition of the ortho-substituted phenyl N-methylThe above discussion suggests that the indicator variable carbamates. In this equation, the govalue is taken as being term in eq 5 is mostly attributed to the effect of the asequivalent to that of corresponding para substituents. The sociation in solvents, l-octanol and water, which is not signs of the p value in eq 6 and 52 are opposite. required to be considered for the correlation with log log I/& = 1.30 (k0.33) n + 2.24 (k0.61) u" Poil-gas. If ethyl ether and methoxyflurane having an oxy base structure are excluded from the set used to derive eq + 1.68 (20.46) HB - 2.56 (k0.29) (52) 5, the slope of the indicator term will probably be reduced a = 1 5 ; r = 0.944; s = 0,223 from 1.9 to 1.6-1.7. The assumption, log poHzo-pas E constant, is only a very We considered that a nucleophilic attack of enzymic active crude approximation. For seven unreactive gases, the serine OH against the side-chain carbamyl-carbonyl "nonliquid" partition coefficients in Table IX are indeed carbon occurs with an assistance of concomitant procorrelated by eq 48. Equations 45 and 48 lead to eq 49 tonation at the carbonyl oxygen for the meta derivatives, suggesting that the susceptibility of log l / p to log Pmt-HzO while without such catalytic assistance for the ortho should be about double of what eq 5 shows. In fact, if we isomers, leading to a common tetrahedral intermediate pick up eight unreactive gases from the set used for eq 5, structure." The slopes of T and HB terms in eq 6 and 52 eq 50 is obtained which conforms to eq 49. By a procedure are quite similar, indicating that the ortho and meta similar to that used in deriving eq 47, we can formulate substituents perhaps fit a common hydrophobic region of eq 51 under conditions of KCH""[solV] >> 1for those having the enzyme and that the ortho and meta basic substituents only one hydrogen-donating site in the molecule. In this probably associate with a common enzymic hydrogen donor by a common mechanism." equation, the terms corresponding to the extra effect are As described earlier, the indicator variable is required asterisked,

log Poct-H20 = log Pooct-H,O t log (1+ KcII"t[octl)/(l

+

Journal of Medicinal Chemistry, 1977, Vol. 20, No. 8 1081

Hydrogen-Bonding Parameter

Figure 7. Stereospecificity in hydrogen bonding with enzymic hydrogen donor.

for 3-acyl, -CN, -NOz, and NMez derivatives in eq 6. 3-Alkoxy compounds are not classified as the hydrogen acceptor. On the contrary, eq 52 requires the indicator variable only of the 2-alkoxy- but not of 2-N02 and -CN compounds. The hydrogen-accepting ability of alkoxy substituents is much lower than the other basic groups as judged by PKHB value of monosubstituted benzenes. Therefore, the enzymic hydrogen donor may be specifically oriented so that it can associate only with alkoxyl groups among ortho basic substituents. With this orientation, its association with meta basic substituents may be controlled not only by a close fitness to the oxo oxygen (depicted as Figure 7) but also by the magnitude of KXEnz-H.For rn-alkoxy compounds the net effect is such that the ratio, KxEnZ“[Enz-H]/Kxoct[oct], becomes close to 1. Since the orientation of the inhibitor molecule in the reversible complex formation with the enzyme is determined by the simultaneous attack of enzymic serine against the carbonyl carbon,“ the association of the basic substituents with the enzymic donor group could be regarded as an “intramolecular” hydrogen-bond formation. The very stereospecific nature of the intramolecular association can be illustrated by a couple of examples. For the equilibrium observed in a,w-diol monomethyl ethers, HO(CH2),,OCH3,by Kuhn and Wires?6 the intramolecular equilibrium constant, Kf(dimensionless),is 12.3 ( n = 2), 1.5 ( n = 3), and 0.67 ( n = 4) in CCll at 20 “C. The association is likely to be enforced specifically by a fivemembered ring association structure in 2-methoxyethanol where the associating partners are located close enough to each other. The Kf values for 3- and 4-methoxyalkanol are similar in magnitude to the intermolecular Kf value (M-’) between alkanol and ether which is around 0.8-1.5.25 Since the enthalpy difference in the intramolecular association does not differ much from that of intermolecular association equilibria,26the larger the magnitude of Kf,the smaller the loss of the translational freedom of intramalecular reactants. The ratio, Kf(intramolecular)/Kr (intermolecular), having a dimension of molarity is supposed to imply the “effective concentration” of the hydroxyl group in the close neighborhood of the alkoxyl group in the intramolecular association equilibri~m.~’The “effective concentration” of the hydroxyl group in 2methoxyethanol is 8- to 15-fold that for the intermolecular association. A similar magnitude of the “effective concentration” is found for intramolecular association of cis-5-hydroxy-2-methyl-1,3-dioxane whose Kf = 18.8 is about the same as that of 2-methoxyethanol considerin the chance factor due to two alkoxy bases in the molecule. These figures of the “effective concentration” for the intramolecular association are quite similar to those anticipated from eq 6 and 52 for the stereospecific association with the enzymic hydrogen donor. The above analyses are thought to rationalize the use of the indicator variable for the hydrogen-bonding effect of substituents in certain cases of QSAR. The present procedure to render a physicochemicalsignificance to the apparent hydrogen-bonding effect appearing as an indicator variable term is to analyze its magnitude in terms of the association equilibrium constants and molarities of

8

hydrogen acceptors (or donors) in bioorganic and reference organic phases. Sometimes the association constants in two organic phases are so similar that they are not important in determining the magnitude of the indicator term. We hope that this procedure will contribute to a better understanding of indicator variables appearing in QSAR, especially in those by means of a multiple indicator variable ap roach currently developed by Hansch and co-workers.

B

References and Notes C. Hansch, J. Med. Chem., 19, l(1976). C. Hansch and E. W. Deutsch, Biochim. Biophys. Acta, 112, 381 (1966). T. Kotani, I. Ichimoto, C. Tatsumi, and T. Fujita, Agric. Biol. Chem., 39, 1131 (1975). T. Higuchi, J. H. Richards, S. S. Davis, A. Kamata, J. P. Hou, M. Nakano, N. I. Nakano, and I. H. Pitman, J. Pharm. Sci., 58,661 (1969); they found a considerable scatter for the correlation of hD with f. D. Clotman, D. van Lerberghe, and Th. Zeegars-Huyskens, Spectrochim. Acta, Part A , 26, 1621 (1970). R. W. Taft, D. Gurka, L. Joris, P. v. R. Schleyer, and J. W. Rakshys, J . Am. Chem. SOC.,91, 4801 (1969). F. Helmer, K. Kiehs, and C. Hansch, Biochemistry, 7,2858 (1968). K. Kiehs, C. Hansch, and L. Moore, Biochemistry, 5,2602 (1966). C. Hansch, A. Vittoria, C. Silipo, and P. Y. C. Jow, J . Med. Chem., 18, 546 (1975). T. Nishioka, T. Fujita, K. Kamoshita, and M. Nakajima, Pestic. Biochem. Physiol., 7, 107 (1977). T. Fujita, K. Kamoshita, T. Nishioka, and M. Nakajima, Agric. Biol. Chem., 38, 1521 (1974). T. Fujita and T. Nishioka, Prog. Phys. Org. Chem., 12,49 (1976). C. Hansch, A. Leo, S. H. Unger, K. H. Kim, D. Nikaitani, and E. J. Lien, J . Med. Chem., 16, 1207 (1973). E. Kutter and C. Hansch, J . Med. Chem., 12, 647 (1969). N. Kakeya, N. Yata, A. Kamada, and M. Aoki, Chem. Pharm. Bull., 17, 2558 (1969). J. H. Hildebrand, J. M. Prausnitz, and R. L. Scott, “Regular and Related Solutions”, Van Nostrand-Reinhold, Princeton, N.J., 1970; A. F. M. Barton, Chem. Reu., 75,731 (1975); T. Wakabayashi, Bull. Chern. SOC.Jpn., 40, 2836 (1967). M. L. Bender and R. B. Homer, J . Org. Chem., 30, 3975 (1965). (a) The [solv] value was estimated from the density of the water-saturated solvent and the molar concentration of saturated water at 25 “C reported in ref 20 and 22. Since the density values of water-saturated ether and cyclohexanol are unavailable, the values for the pure liquid were used. For these two solvent phases, the [solv] value may be slightly underestimated. (b) J. H. Lady and K. B. Whetsel, J . Phys. Chem., 71, 1421 (1967). T. Gramstad and 0. Vikane, Spectrochim. Acta, Part A , 28, 2131 (1972). R. N. Smith, C. Hansch, and M. M. Ames, J . Pharm. Sci., 64, 599 (1975). R. Collander, Acta Chem. Scand., 3, 717 (1949); 4, 1085 (1950); 5, 774 (1951). A. Leo and C. Hansch, J . Org. Chem., 36, 1539 (1971). P. Seiler, Eur. J. Med. Chem., 9, 473 (1974). K. W. Miller, W. D. M. Paton, E. B. Smith, and R. S. Smith, Anesthesiology, 36, 339 (1972). M. D. Joesten and L. J. Schaad, “Hydrogen Bonding”, Marcel Dekker, New York, N.Y., 1974, p 293. L. P. Kuhn and R. A. Wires, J . Am. Chem. Soc., 86,2161 (1964). W. P. Jencks, “Catalysis in Chemistry and Enzymology”, McGraw-Hill, New York, N.Y., 1969, p 10. G. Aksnes and P. Albriktsen, Acta Chem. Scand., 20,1330 (1966). M. Yoshimoto and C. Hansch, J. Med. Chem., 19,71 (1976).